Evolving in Thought Space: Training a Small Model at Test Time Unlocks Better Discoveries
cs.AI, cs.LG
Submitted: 2026-10-05
Updated: 2026-10-12
Code: https://github.com/Human-Agent-Society/reef
Sources
- Aster: Autonomous Scientific Discovery over 20x Faster Than Existing Methods
- K-Search: LLM Kernel Generation via Co-Evolving Intrinsic World Model
- GLM-5: from Vibe Coding to Agentic Engineering
- FrontierCS: Evolving Challenges for Evolving Intelligence
- AlphaEvolve: A coding agent for scientific and algorithmic discovery
- CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery
- Credit Assignment with Resets in Language Model Reasoning
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Towards Execution-Grounded Automated AI Research
- Learning to (Learn at Test Time)
- Algorithm Discovery With LLMs: Evolutionary Search Meets Reinforcement Learning
- ThetaEvolve: Test-time Learning on Open Problems
- AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation
- Qwen3 Technical Report
- Structured Scaling of AI Discovery Across Diverse Scientific Domains
- Learning to Discover at Test Time
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems